The rapid proliferation of short videos on social media platforms has led to a significant increase in derivative content, often repurposing the same video segments. In response to this phenomenon, our study specifica...
详细信息
ISBN:
(纸本)9798350351439;9798350351422
The rapid proliferation of short videos on social media platforms has led to a significant increase in derivative content, often repurposing the same video segments. In response to this phenomenon, our study specifically addresses short video formats, developing an innovative methodology to swiftly identify content that is either similar or identical. Recognizing the inefficiency inherent in comparing the entirety of visual content, our approach initially involves the dimensionality reduction and categorization of short videos through textual data. For the analysis of visual data, we implemented a two-fold strategy: initially removing the template and segmenting videos in the preprocessing phase, followed by the application of two distinct analytical models, ViSiL and CLIP. These models are employed to detect reused materials across a variety of contexts. Our empirical findings revealed multiple instances of footage repetition. This research contributes to two main aspects: (a) analyzing the effectiveness of additional video information for clustering, and (b) developing a more efficient method for detecting identical source materials.
暂无评论